Season 3: Episode #92
Podcast with Matthew Roman, Chief Digital Strategy Officer, Duke University Health System
In this episode, Matthew Roman discusses how Duke Health is implementing a number of foundational technology platforms for effective patient engagement and care delivery over the next couple of years.
A clinician by background, Matthew describes the collaboration model among a diverse group of technology and operational executives to implement digital health programs at Duke Health. He gives us a hint of the one single question he wrestles with every day as the Chief Digital Strategy Officer. He also explains why they choose to “tread lightly” in offering clinical advice through artificial intelligence.
Matthew describes several challenges digital health startups must be prepared to face, even if they have remarkable and game-changing technology solutions. Among his words of advice? Don’t oversell. He also shares a few learnings from his experience for peer group executives in health systems. Take a listen.
|01:17||About Duke Health and the patient populations.|
|02:34||Tell us a bit about the digital programs currently operational at Duke Health, maybe touch on telehealth in particular.|
|05:07||Talk to us about your top foundational platforms, any ones that you used to execute, and also your whole strategy. Are you using one or you are using multiple platforms for different things? How do they all fit with your other tools, especially the EHR platform? programs?|
|07:25||Have you been using chatbots more in the context of clinical chats or more in an administrative context for enabling access and providing patients with information on self-service tools?|
|09:27||Where you are in your CRM journey and what your focus areas are with the CRM platform?|
|12:09||In the context of a chronic disease, most of the deployments have been from RPM standpoint. How is it worked so far, especially the aspects where you bring back the data from the devices and the sensors and you try to combine that with the patient longitudinal records in the EHR?|
|16:11||How are you driving data and analytics program at Duke Health and how you are harnessing emerging data sources and tools such as AI?|
|18:12||How are you structured to drive digital strategy? can you share your org model for driving digital transformation?|
|23:29||What is your advice for tech firms, especially startups and innovators who want to be a part of your digital journey?|
|25:49||Can you share a couple of best practices and operating principles for success with digital health programs?|
Q: Tell us a bit about the populations you serve at Duke Health and your role in the organization.
Matthew: I’m the Chief Digital Strategy Officer for Duke University Health System. It is a medium-sized yet a very high-quality academic medical center located in the center of North Carolina. We’re pretty proud of the quality of care we offer through our three hospitals — a flagship academic hospital and two community hospitals — along with a large series of clinics, both primary care and a large specialty faculty practice. I report to the CIO and we support the academic mission through the Schools of Medicine and Nursing, as well as the health system functions.
Q: Can you share an overview for some of the digital programs currently in-flight at Duke Health? Telehealth, for instance, has been a big growth area for most organizations. Which one has it been for you?
Matthew: Our Digital Strategy Office was formed about three-and-a-half years ago as envisioned by our CIO, a physician himself. We are responsible for consumer-friendly, patient-facing technologies to help with our patients’ attempts to engage with us as a health system. We’re deep in the throes of implementing a number of, what I would call, foundational technology platforms on which, over the next couple of years, we will build hopefully more effective and broader reaching use cases. So, these platforms include programs, some of which are fully embedded already, some of which are in-flight.
Through our telehealth platform — our patient portal – we are trying to improve patients’ experiences. A CRM strategy around conversational AI and chatbots does exist but it’s important to reach out to the patients to learn from them what they want from us. We’re doing this through a virtual Patient Advisory Council. Some others have done this as well along with remote patient monitoring, both, in support of the telehealth platform and both supporting continuing care via virtual visits. Even if that care is initially delivered in-person, we’re able to — through these remote patient monitoring strategies — capture data points in much greater frequency to support clinical decision-making and predictive modeling.
Q: How have your patients and caregivers responded to Telehealth? What were your platforms and strategy for execution? How do they align with your EHR and other tools involved in delivering a seamless experience to patients?
Matthew: Our experience was like most others. We had a pretty small telehealth footprint. We had some early adopters and really impressive work, pre-pandemic, like our Movement Disorder Clinic. It had a Neurologist who was a very early adopter of telehealth. His patients were A-listers with tremendous movement and mobility disorders, and it took an army to bring them to our clinic. He had a pretty wide capture rate or geography and so, we were able to work with him to enable video visits to these patients. We had the same hockey stick increase in volume as everybody else did in March 2020. We went from 100 visits a month to 2000 visits a day, much like everybody else. The truth is, our highest month volume since the start of the pandemic was March 2021 and then, we’ve started to tail off just a little bit. We continue to have pretty high volume in some specialties or behavioral health and psychiatry clinics have remained very high adopters and high utilizers of our primary care clinics and certainly some of the specialty and surgery clinics as well. We have a primary platform that’s embedded in our EHR. And we have a backup platform, too. This way we’re able to capture patients even if they don’t have an app on their devices or face connectivity issues. Then, we can rescue or salvage that by sending a rescue link. We have two active platforms that we’re working with currently.
Q: You also mentioned the chatbots. Have you used them more in the context of clinical chats or in the administrative context to enable access and provide patients with information on self-service tools? Or are you doing both?
Matthew: This is a great question! We’re in the relatively early stages of implementing our chatbot and we’re cutting our teeth on administrative functions. We will tread lightly in offering clinical advice through AI, more from risk tolerance and quality assurance perspectives than anything else. I think that we’re starting from an administrative place to access some instructions, directions, wayfinding, touchless arrival, etc., and then, we’ll branch from there.
Q: Is your approach to start small, establish adoption levels and make sure that the chat works effectively and people feel comfortable before you get to the more complex, high stakes, high-risk kind of functions?
Matthew: That’s right. We’re also working hard with these platforms but the connection between them is what’s really so intriguing to me. For instance, if the patient had a remote monitoring device at home or when monitoring their BP via home checks, they engage with us via chatbots, our response is informed by the fact that the patient is being monitored. So, we could be smarter in our response and answer the patient differently via AI.
Q: Where are you in your CRM journey? What are your focus areas with the CRM platform?
Matthew: We’ve implemented an enterprise level CRM in our marketing strategy. So, that was our first stretch into CRM many years ago. Since then, at our Duke Clinical Research Organization, a large CRO, we have an installation of the same CRM tool that helps manage multicenter trials, not just site-based research into a bunch of work in the CRM but in the unit, too. Now, we’re in the 11th hour of our implementation of the CRM tool and our Access Services Center with its multiple hubs to serve our primary and specialty care providers.
What we are hoping to do is get a little smarter in our engagement, knowing who the patients are, who’s calling and their call history, which right now we don’t have much insight into but which we’ll be able to add this year. I can, very easily, envision patient acquisition thus.
From the marketing effort within the CRM tool and creating journeys from the time we acquire a patient to when we actually schedule that patient for the needed/requested services to then linking them to the portal and other things that we have downstream to continuously push engagement — clinical and administrative – so as to reduce friction and lower the barrier for entry.
Q: With regard to the last foundational platform — remote monitoring — in the context of a chronic disease, where most of the deployments have been from RPM standpoint, how has it worked thus far? Any learnings you’d like to share?
Matthew: We’re taking an approach that RPM has two really big buckets. The first bucket is to replicate what’s happened over generations. When we walk into our provider’s office, we get weighed, core temperature, height, blood pressure and heart rate measured so we can replicate that when the virtual visit occurs by remote capture, just to continue to be able to capture the same sort of quality data that we have for generations. More importantly, though, this is so that when our providers are being asked to make clinical decisions based on a single data point or very precious few data points over a long longitudinal time point, then we don’t make under-informed decisions.
We’re structuring it so we send patients home with whatever the appropriate biometric kit might be — be it blood pressure, glucose monitoring or pulse oximeter, etc. Bring these data into a lake or a repository short of our EHR where we can analyze the data and apply rules to trigger alerts. These will be alerts to the provider and care teams. If there’s a either a series of or a sequence of progressively out of range numbers or an alert or a value that’s particularly high or low, that’s somewhat dangerous and we may want to intervene or send alerts to patients.
And these might be an alert to patients because we haven’t received a value in a few days or because the values are trending well and we want to send them a nudge that says -“Congratulations! Good job! The work you’re doing is effective and your blood pressure is becoming under control. You’ve lost five pounds or the reverse.”
If the trends are actually going in the wrong direction, we want to send encouraging messages to help them get back on course and nudge the provider to maybe change the course in one way or another. The long game is once we capture enough of these data points across a broad enough segment of our population, it’s representative enough. Then, we’ll get smart about what normal recovery looks like after a procedure and know what normal or well looks like when a variant in the data is actually meaningful or when it’s a predictable variant that’s innocuous.
We’re blessed to have really tremendous data science people around to look at these big data sets and find the pearls. So, we’ll be able to set up predictive models to understand when data are mandating action be taken. This also has, workflow utility, because it can help us give patients a heads-up on events, they can expect to occur somewhere between day 4- 6 after they are back home. So, there’s some workflow utility as well. That’s our journey.
Q: How are you driving data and analytics? How are you set up to serve the multiple needs of the enterprise? What do your structures and successes look like?
Matthew: I’m a consumer of these very brilliant people I work with. One of my peers is the Chief Analytics Officer, whose team’s responsible for all the structure. They’ll explain this better but remember the lake I mentioned earlier? That was built for us to pile-in multiple data-streams. In the near future, we may make informed clinical decisions based on things beyond just the very rich EHR data. That alone is incomplete, of course. So, in this lake or repository we’ll have RPM data, social determinate data, expense/spends as well as location data to facilitate our remote monitoring journey. All of this, of course, with the consent of patients for they will be its greatest beneficiaries.
Q: Your role reports to the CIO. What’s the organization model for driving digital transformation? How did that start?
Matthew: While I report to the CIO, I’m not a deep technician nor an engineer. I come at this from a clinical angle because I’m a clinician, first, and a strategist and digital health person, second. I have a small but diverse team with broad backgrounds — from clinical informatics to physical therapists — including a nurse and a physician, who’s our Medical Director that’s responsible for our portal.
We work very carefully and closely with our colleagues in the health system — clinical and operational leads — to understand the opportunities that our clinicians can have. Our budget is also through our IT shop so we do try to make clever use of technology to ease workflows and enhance abilities of clinicians to engage with patients and empower them with information and tools to supplement their care between clinical encounters.
Our operational colleagues are critical cogs in this wheel that help implement workflows, set appropriate impact metrics, have baseline days against which to compare. I call them impact metrics because it’s not just about numbers of adoption on our portal account; it’s to understand what difference we may have made.
Q: How do you approach technology choices for your transformation especially when it comes to the risks?
Matthew: That’s a question that, candidly, I wrestle with every single day. We have invested significantly in our EHRs – both, dollars and effort. We have a very mature installation of enterprise EHR but it’s our transaction tracking and our medico-legal record keeping system. And that’s important.
We work hard because our clinicians are extremely busy people. In keeping with a concept shared with me by our previous CTO — a classic single pane of glass – I must say we have a fairly high bar in the EHR; high enough for us to tell our staffers, clinicians and administrators to go to another application for a particular purpose. When we want to bring in another application, we try to allow us to be able to launch it from within the primary health record, the place where our staff are working. We insist on single-sign-on, being able to preserve contextual awareness. So, our pendulum swings all the time between high level enterprise solutions and fit for purpose. And it’s an internal struggle. All this is to say that I know I’m not answering your question clearly, but it is maybe the unanswerable one.
Q: When it comes to innovation and innovative technologies, how do you parse through all that’s happening now in the market to find that little nugget that will stand the test of time?
Matthew: With startups, some of this advice is welcome. However, for a complex organization like ours, the sales cycle is longer than you, the startup or I would like it to be, but it’s just the reality. We work very hard to shorten it, but it’s complex. I’m not saying that’s right, but it just takes a long time. So, be patient.
I think that the point that we made a moment ago about respecting the single pane of glass as much as possible is important, even if that widget is just simply remarkable and game-changing. If we can get it in front of the users and the best clinicians, the patients, then it won’t matter.
In other words, there is a tipping point where we can put too many applications on a patient’s device and then it becomes noise rather than signal. For a patient who has comorbid conditions — and we have three or four really magical applications that could change that patient’s course if we could elegantly get that patient to interact with that application — it’s somewhat meaningless.
So, the integration of patience and single pane of glass should be easier over time because of FHIR standards, smart application capabilities in these sorts of things and the underselling and over-delivery. If it’s a niche product, it’s what it is. But the other side of that continuum is the large company or the or the medium-sized company who comes into an organization like this one and says we can solve all your problems. That’s somewhat of an oversell.
Q: That’s good advice. You’ve been in the role for a few years so you’ve had success and times when things didn’t go your way. What’s your advice or best practices that you would like to share with your peers on similar transformations?
Matthew: I love the question and I would answer it by saying — be persistent, tenacious and don’t stop. I won’t tell you that I have better practices because this is a personal semantic question for me. I don’t like that term because it implies that I already have what’s best and it can’t get better. To me, the answer is tenacity.
Try something carefully, monitor the impact, make a change, try something again. That, I happen to think, is the key. Don’t be afraid to try something new, be obviously cautious and judicious in these changes because we’re talking about patient safety. But where possible, the classic fail-fast mentality to me is wise. And then once you’ve failed, you change, learn and reapply,
About our guest
Matt Roman serves as the Chief Digital Strategy Officer for Duke University Health System. He is responsible for developing and deploying consumer-focused digital strategies, implementing innovative technologies to better engage patients and families, and extending our health IT footprint out into the community. Matt is passionate about building an optimal care experience for patients, so they can maximally engage in their health and wellness during and between clinical encounters. As a clinician himself, Matt is empathetic to needs of providers and strives to improve efficiency in care delivery while also improving clinical outcomes and supporting research.
Matt’s teams are responsible for initiatives to include digital health, remote patient monitoring, CRM deployment, patient experience, the patient portal, and utilization of conversational AI in enhancing patient experience, among other strategic initiatives.
Matt has extensive experience in hospital and clinic operations. He ran the enterprise command centers during the health system’s electronic health record go-live, partnered with clinical and operational leaders to establish enterprise IT governance, and worked closely with community leaders to bring our EHR to non-Duke clinics like our local FQHC, Lincoln Community Health Center. Matt has partnered with the clinical community to optimize clinical workflows and maximize the utility of our EHR for busy clinicians. Matt is responsible for designing and deploying technologies to support patients through their health care journey and for working with providers and health system leadership to derive maximal value from our investments in health information technology.
About the host
Paddy is the co-author of Healthcare Digital Transformation – How Consumerism, Technology and Pandemic are Accelerating the Future (Taylor & Francis, Aug 2020), along with Edward W. Marx. Paddy is also the author of the best-selling book The Big Unlock – Harnessing Data and Growing Digital Health Businesses in a Value-based Care Era (Archway Publishing, 2017). He is the host of the highly subscribed The Big Unlock podcast on digital transformation in healthcare featuring C-level executives from the healthcare and technology sectors. He is widely published and has a by-lined column in CIO Magazine and other respected industry publications.